Santosh,
-----Original Message----- From: Santosh P K [mailto:[email protected]] Sent: Tuesday, June 30, 2015 10:27 AM To: Shahram Davari; Gregory Mirsky; Vengada Prasad Govindan (venggovi) Cc: [email protected] Subject: RE: New Version Notification for draft-spallagatti-bfd-vxlan-00.txt Shahram, > Assuming you want to do VTEP-to-VTEP connectivity, The only way that I > think this draft can provide extra information compared to running BFD over > the outer IP tunnel, is when the Inner MAC-DA is the MAC-DA of the VTEP > port that connects to the VM. In such case the packet will exercise the > VXLAN/VNI forwarding (DA, VXLAN/VNI) and then the inner L2 is terminated > at the VTEP port connecting VTEP to VM and BFD is extracted and processed. > Basically some kind of UP-MEP on the VTEP. When you say that inner DA-MAC should be that of the MAC connecting to VM, you mean the other side of the interface and not the receiving interface on VTEP right? SD>Correct. Doing so ensures VXLAN/VNI is exercised. Otherwise the VXLAN is just a meta-data and it can be carried inside BFD packet (such as adding a TLV). In fact you can create a BFD TLV and add many VXLANs to it and then a single BFD session can test multiple VXLANs that start and terminate on same VTEPs. SD> Assume there are 10 VXLAN/VNI between VTEP1 and VTEP2. Based on your draft you want to run 1xBFD per VXLAN/VNI. So you have 10xBFD sessions in each direction. My question is how are these BFD sessions different from each other? In other words is there a case that some of these BFDs show connectivity and some don't? SD> Instead, why don't you run just 1 BFD session and list all VNI/VXLANs inside the BFD as a list. Basically aggregate. Since you are not using VNI/VXLAN for any forwarding. You are just using it for demultipexing. When we are verifying VTEP to VTEP connectivity packet has to be absorbed by VTEP itself and when it does it needs to demux BFD packet with some key and that would be VNI in the VXLAN header. This is what we are talking about in the draft, maybe we are much more clear in the next version of the draft which will be published soon. SD> RFC 5881 supports what you need. This RFC applies to any Tunnel: " This application of BFD can be used by any pair of systems communicating via IPv4 and/or IPv6 across a single IP hop that is associated with an incoming interface. This includes, but is not limited to, physical media, virtual circuits, and tunnels" Thx Shahram Thanks Santosh P K > > -----Original Message----- > From: Rtg-bfd [mailto:[email protected]] On Behalf Of Shahram > Davari > Sent: Tuesday, June 30, 2015 9:23 AM > To: Santosh P K; Gregory Mirsky; Vengada Prasad Govindan (venggovi) > Cc: [email protected] > Subject: RE: New Version Notification for draft-spallagatti-bfd-vxlan-00.txt > > Santosh, > > But your draft says: > > " BFD session can be used to run between VM's for VM's connectivity check:" > > So, which one is it? VTEP-to-VTEP or VM- to-VM? > > Also what is the MAC-DA of inner Ethernet header? > > Thx > Shahram > > -----Original Message----- > From: Santosh P K [mailto:[email protected]] > Sent: Tuesday, June 30, 2015 2:13 AM > To: Shahram Davari; Gregory Mirsky; Vengada Prasad Govindan (venggovi) > Cc: [email protected] > Subject: RE: New Version Notification for draft-spallagatti-bfd-vxlan-00.txt > > > > Another question I have is how do you forward these BFD packets from > > VTEP to VM? Since DIP = 127/8, I assume there is no way to forward > > such packets past the VTEP. > > We don't want to send it past VTEP as BFD will terminate at VTEP itself. Right > now use of 127/8 is not really clear as inner IP address. We have again added > some text for inner IP address and its use. > > > > > > If you want to do VM to VM connectivity check then I suggest one of > > the > > following: > > > > 1) if VMs are only L2 aware and VXLAN is L2VPN, the run Ethernet OAM > > between VMs > > 2) If VMs are L2 and L3 aware and VXLAN is L2VPN then run BFD over > > IP/UDP over Ethernet over VXLAN > > 3) If VMs are L3 aware and VXLAN is L3VPN then you have 2 cases: > > 3a) If VTEP and VMs are physically in the same CPU core > > then run what you are proposing in this draft (1-hop BFD). Although it > > should give you same result as running BFD over the outer IP tunnel > > 3b) If VTEP and VMs are physically separate (such as VTEP is > > in TOR and VM is CPU core), then run something similar to what you are > > proposing but with Multi-hop DIP address so that BFD can be forwarded to > VM from VTEP. > > > > We are not trying to address VM to VM connectivity check as I believe that > does not really need any changes and BFD as it is should run. Proposal is from > VTEP to VTEP. > > > Thanks > Santosh P K > > > -----Original Message----- > > From: Rtg-bfd [mailto:[email protected]] On Behalf Of Shahram > > Davari > > Sent: Monday, June 29, 2015 4:15 PM > > To: Gregory Mirsky; Vengada Prasad Govindan (venggovi) > > Cc: [email protected] > > Subject: RE: New Version Notification for > > draft-spallagatti-bfd-vxlan-00.txt > > > > Hi Greg, > > > > I don't see much value in running BFD for SS-PW compared to running > > BFD over the LSP. > > > > Thx > > SD > > > > -----Original Message----- > > From: Gregory Mirsky [mailto:[email protected]] > > Sent: Monday, June 29, 2015 2:21 PM > > To: Shahram Davari; Vengada Prasad Govindan (venggovi) > > Cc: MALLIK MUDIGONDA (mmudigon); Santosh P K; [email protected] > > Subject: RE: New Version Notification for > > draft-spallagatti-bfd-vxlan-00.txt > > > > Hi Shahram, > > I'll get to VXLAN shorty but wanted to ask quick question about SS-PW. > > True, we have case of MS-PW where PW label used (that was discussed in > > MPLS- TP in particular). But that doesn't mean that VCCV BFD has value > > only for MS- PW and for SS-PW has no value. Would you agree? > > > > Regards, > > Greg > > > > -----Original Message----- > > From: Shahram Davari [mailto:[email protected]] > > Sent: Monday, June 29, 2015 2:02 PM > > To: Gregory Mirsky; Vengada Prasad Govindan (venggovi) > > Cc: MALLIK MUDIGONDA (mmudigon); Santosh P K; [email protected] > > Subject: RE: New Version Notification for > > draft-spallagatti-bfd-vxlan-00.txt > > > > Hi Greg, > > > > You are welcome. Could you please clarify which one of the following > > is the packet format: > > > > 1) OuterL2-Outer IP-UDP-VXLAN-Inner L2-Inner IP-UDP-BFD > > > > Or > > > > 2) OuterL2-Outer IP-UDP-VXLAN-Inner IP-UDP-BFD > > > > If (2) hen how do you specify the VXLAN payload in IP and not Ethernet? > > > > Also This is different from PW BFD, since in case of PW, there can be > > MS-PW, where the LSP BFD is not end-to-end. But in this case we don't > > have MS- VXLAN. > > So the span of the VXLAN and the IP tunnel is the same. > > > > In other words you have to specify in which part of forwarding the BFD > > the VXLAN Tag is used. If it is not used then it has no effect. > > > > Thx > > Shahram > > > > -----Original Message----- > > From: Gregory Mirsky [mailto:[email protected]] > > Sent: Monday, June 29, 2015 1:51 PM > > To: Shahram Davari; Vengada Prasad Govindan (venggovi) > > Cc: MALLIK MUDIGONDA (mmudigon); Santosh P K; [email protected] > > Subject: RE: New Version Notification for > > draft-spallagatti-bfd-vxlan-00.txt > > > > Hi Shahram, > > thank you for your comments to this proposal that make the discussion > > so much alive. > > I think that processing of the VXLAN tag by VTEP before validating BFD > > is sufficient, in my view, level of verification. In VCCV BFD the PW > > label is not used for BFD forwarding but we find it useful as Service > > OAM in addition to RFC 5884, BFD over MPLS LSP. > > > > Regards, > > Greg > > > > -----Original Message----- > > From: Shahram Davari [mailto:[email protected]] > > Sent: Monday, June 29, 2015 10:39 AM > > To: Vengada Prasad Govindan (venggovi) > > Cc: Gregory Mirsky; MALLIK MUDIGONDA (mmudigon); Santosh P K; rtg- > > [email protected] > > Subject: RE: New Version Notification for > > draft-spallagatti-bfd-vxlan-00.txt > > > > Hi Prasad, > > > > Is this a special type of BFD or standard BFD RFC 5880 and 5881)? > > Since standard BFD processing does no care where the BFD came from it > > only looks at "your discriminator" to update BFD state machine. > > > > Also I don't see how many VXLANs can be checked via a single BFD session. > > Could you please describe? > > > > Lastly checking to see a VXLAN/VNI forwarding domain exist in a VTEP > > should not require BFD. Just use some query mechanism. Why do you > need > > to run continuous BFD. > > > > What you have to show me to convince me that your draft solves a real > > problem is to show that VXLAN tag is used for BFD forwarding. > > Otherwise BFD over the outer or Inner IP should give you all coverage > needed. > > > > > > Thx > > Shahram > > > > > > > > > > -----Original Message----- > > From: Vengada Prasad Govindan (venggovi) [mailto:[email protected]] > > Sent: Monday, June 29, 2015 3:11 AM > > To: Shahram Davari > > Cc: Gregory Mirsky; MALLIK MUDIGONDA (mmudigon); Santosh P K; rtg- > > [email protected] > > Subject: RE: New Version Notification for > > draft-spallagatti-bfd-vxlan-00.txt > > > > Hello Shahram, > > At the terminating VTEP, VxLAN information is used to consume the > > BFD packet. In other words, a BFD session increases the confidence of > > the existence of the VNI-Forwarding Domain mapping and the presence of > > valid VTEP termination configuration at the terminating VTEP. At the > > originating VTEP, it is a matter of implementation of how many VxLAN > > tables are exercised in the datapath (am aware of at least one > > implementation where it is being exercised to a considerable extent). > > > > Thanks > > Prasad > > > > -----Original Message----- > > From: Shahram Davari [mailto:[email protected]] > > Sent: Saturday, June 27, 2015 8:24 PM > > To: Shahram Davari > > Cc: Vengada Prasad Govindan (venggovi); Gregory Mirsky; MALLIK > > MUDIGONDA (mmudigon); Santosh P K; [email protected] > > Subject: Re: New Version Notification for > > draft-spallagatti-bfd-vxlan-00.txt > > > > Hi > > > > May be a better way to make this clear is to answer the following question: > > > > Where is the VXLAN tag information used in this BFD forwarding? > > > > Thx > > Shahram
